arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2108.01285cs.LGcs.CV
收藏

空间无偏生成模型

Toward Spatially Unbiased Generative Models

Seoul National University(首尔国立大学);Samsung SDS(三星SDS)

Jooyoung Choi, Jungbeom Lee, Yonghyun Jeong, Sungroh Yoon

摘要:

最近的图像生成模型显示出显着的生成性能。然而,它们反映了数据集中强烈的位置偏好,我们称之为空间偏差。因此,生成器在不可见的位置和比例处渲染较差的采样。我们认为,生成器依赖于他们的隐式位置编码渲染空间内容。根据我们的观察,生成器的隐式位置编码是可变的,使得生成器在空间上有偏差。为了解决这个问题,我们建议在生成器的每个尺度上注入显式位置编码。通过学习空间无偏生成器,我们促进了在多个任务中的生成器的鲁棒使用,例如GAN反演、多尺度生成、任意尺寸和纵横比的生成。此外,我们表明,我们的方法也可以适用于去噪扩散概率模型。

英文摘要:

Recent image generation models show remarkable generation performance. However, they mirror strong location preference in datasets, which we call spatial bias. Therefore, generators render poor samples at unseen locations and scales. We argue that the generators rely on their implicit positional encoding to render spatial content. From our observations, the generator's implicit positional encoding is translation-variant, making the generator spatially biased. To address this issue, we propose injecting explicit positional encoding at each scale of the generator. By learning the spatially unbiased generator, we facilitate the robust use of generators in multiple tasks, such as GAN inversion, multi-scale generation, generation of arbitrary sizes and aspect ratios. Furthermore, we show that our method can also be applied to denoising diffusion probabilistic models.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑